On Sun, Oct 31, 2004 at 03:36:54PM -0500, Dheeraj Reddy wrote:
> On Sun, Oct 31, 2004 at 02:28:22PM -0600, David Young wrote:
> 
> > > 
> > > I still see an issue with it: if the driver *knows* what are the correct
> > > units, why shouldn't it provide that information to userland, instead of
> > > requiring me to calibrate the card?
> > 
> > What with manufacturing variation, aging components, and the variety of
> > power amps, how does a driver know the correct units?  I feel fairly
> > certain that there is not a driver in NetBSD that can tell us with
> > what its transmit power units are with the certainty we need for, say,
> > complying with wireless regulations.
> > 
> 
> We could standardize and do all driver-independent calculations in mW
> and let all drivers expect to be passed mW and then they could act
> according to the units which the respective firmware/card expects (
> converting mW to dB or any other unitless factor that they think is
> appropriate) ? Of course, I am ssuming here that the device documentaion
> contains relevant information about units.

Documentation doesn't ordinarily reveal that information.  Given the
considerations above, I don't think it would matter too much if it did.

I maintain that if you have compliance in mind, or some scientific
purpose, the driver will not provide reliable calibration information, so
it's not worth it to put "smarts" into the kernel to try to extract it.
Furthermore, if a user does not have compliance or research in mind,
then unitless settings are quite sufficient.

Dave

-- 
David Young             OJC Technologies
dyoung@ojctech.com      Urbana, IL * (217) 278-3933